{"id":"2e4665f9-6a75-4b37-b3c7-56419097b82f","arxiv_id":"1908.02946","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Truebit can pay verifiers with minted TRU tokens while task givers buy fixed CPU tokens, giving computation tasks a stable on-chain price without external oracles.","lead":"This paper outlines a token economy for the Truebit computation network where tasks are always priced at one stable CPU token per computation step and rewards are minted in a separate TRU token. The design aims to bootstrap the network without external price oracles, exchanges, or trusted nodes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sustainability Proposition's x < p condition is undermined by the protocol's own conversion rate: Strategy 2 lowers the median price, which raises the CPU minting rate and can push x above p.","rationale":"The paper's stated goal is a stable task-pricing token whose economics are sustainable without external oracles or exchanges. The one condition that has to be true for that claim is the Proposition in §3.2: the Strategy-2 hodlers must eventually absorb the token supply, which requires the per-epoch creation rate x to stay below the hodler fraction p. The reader flagged that p and x are asserted, not derived. I agree, and I think the gap is even more specific: the tasking conversion contract ties x to the median price, and Strategy 2 acts precisely to lower that median. The protocol therefore contains a feedback loop that can push x toward or above p. This makes the Proposition's assumption an internal consistency question, not just a behavioral one. The exchange-arbitrage Claim also assumes a fixed conversion rate, which is the same endogenous median, so it cannot be invoked to stabilize the system. A concrete numerical or analytical test of the coupled median-price/token-creation dynamics would settle whether any parameter regime satisfies x < p while q > 1/2. If the test finds such a regime, the conditional verdict stands and the paper's design sketch becomes more credible; if not, the sustainability claim should be rejected. I therefore keep the reader's CONDITIONAL verdict: the concern is significant but addressable by a focused analysis, and the paper is a design proposal rather than a claimed formal proof.","tokens_in":11221,"tokens_out":23149,"duration_ms":249569,"concrete_test":"Model the coupled dynamics: let p_t be the median local price and C_t the TRU volume burned in the tasking conversion contract per epoch; then x_t = C_t / p_t. Postulate a Strategy-2 share q_t > 1/2 posting p_t·0.8 and others posting p_t·1.2, so p_{t+1} is the median of the resulting distribution. Numerically search the (p_0, C_0, elasticity of C to p, q_0) parameter space for a trajectory with q_t > 1/2 and x_t < q_t for all t. If no such trajectory exists, the Proposition's hypothesis is incompatible with the protocol's conversion rule.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2's central sustainability result is the Proposition, whose convergence time n/(p−x) requires x < p, where p is the fraction of Solvers/Verifiers following Strategy 2 and x is the per-epoch creation rate of new CPU tokens. The paper calls these 'reasonable conditions' and never derives them from individual rationality or from the protocol's mechanics. This is not merely an unproven behavioral assumption. The tasking conversion contract described in the same section mints CPU at rate 1/p_med, where p_med is the median local price. Strategy 2 is defined as posting a local price near the median minus 20%, and if more than half of participants do this, p_med falls. As p_med falls, the same volume of TRU burned in the conversion contract produces more CPU tokens, so the token creation rate x = C/p_med rises. The Proposition's inequality x < p is therefore not an exogenous 'reasonable condition'; the protocol's own dynamics push x upward precisely when the Strategy-2 coalition succeeds. If the conversion volume C does not shrink fast enough, x crosses p and the geometric-series argument diverges. The exchange-arbitrage Claim has a related gap: it assumes a fixed conversion rate r, but r is the endogenous median price. Without a bound on C or a proof that the median-price dynamics keep x < p, the sustainable-economics claim is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a two-token design for Truebit: a stable 'CPU' token used for task payments at a fixed rate of one CPU per computation step, and a mintable 'TRU' reward token paid to Solvers and Verifiers. Solvers/Verifiers post local prices in TRU per step; rewards are minted at the median of these local prices, and TRU can be converted into CPU through a tasking conversion contract at the same median price, policed by Staking Monitors and Tasking Monitors. The paper argues that this median-based pricing, together with a 'Strategy 2' hodling behavior, yields sustainable economics through a geometric convergence argument, and that external exchange markets do not disrupt the construction under a fixed conversion rate. It then sketches bootstrapping via external token staking, a DAO governance layer that dissolves into TRU/CPU, an upgrade mechanism, and two protocol modifications (a martingale defense and random selection of Verifiers).","tokens_in":11470,"tokens_out":5168,"duration_ms":60534,"significance":"If the economic claims were established, the paper would offer a coherent, oracle-free mechanism for stable task pricing and for bootstrapping a new token system from existing liquid assets. The CPU token's task-price stability is definitional (one token per computation step), the staking-monitor mechanism gives a concrete check on local price reporting, and the geometric-series argument is transparent under its stated assumptions. The governance-dissolution idea is original. However, the central sustainability claim is not yet supported: the Proposition in Section 3.2 assumes the behavioral dominance it aims to establish, and its convergence condition x < p is coupled to the protocol's own median-conversion dynamics. The exchange-stability Claim likewise treats the conversion rate as fixed when it is endogenous. These are load-bearing issues for the paper's main economic thesis, so the significance is conditional on substantial revision.","major_comments":[{"comment":"The Proposition assumes (a) at least p fraction of Solvers/Verifiers follow Strategy 2 and (b) on average x < p new CPU tokens are created per epoch, and then concludes convergence in n/(p - x) tasks. Assumption (a) is precisely the behavioral dominance claim ('Strategy 2 is the long-run, dominant strategy for rational miners') that the surrounding text says it will describe but never derives from individual rationality; no best-response or equilibrium analysis is given. Assumption (b) is not an exogenous 'reasonable condition' either: the tasking conversion contract mints CPU at rate 1/p_med, and Strategy 2 is defined as posting a local price near the median minus 20%, which pushes p_med down and therefore pushes the minting rate C/p_med up. Unless conversion volume C is bounded or a fixed-point argument establishes x < p, the geometric-series convergence is unsupported. This undermines the abstract and Section 3.2 claim of 'sustainable economics.'","section":"Section 3.2, Proposition"},{"comment":"The arbitrage Claim assumes 'a fixed tasking conversion rate r,' but r is the median of bonded local prices and is endogenous to the strategy mix and conversion volume. The Claim therefore does not establish that USD(CPU) tends to r * USD(TRU) unless the dynamics of r are modeled. As written, this argument is conditional on the same missing median-price dynamics as the Proposition, and it cannot be invoked to show that external exchanges do not disrupt the construction.","section":"Section 3.2, Claim on exchange markets"},{"comment":"The design principle that 'the value of tokens paid into the Truebit protocol must not exceed the value of tokens paid out as rewards' is not reconciled with the on-the-fly minting of TRU rewards described in the same section. If rewards are minted at the median local price, the quantity of TRU a Task Giver can obtain by issuing a private task and burning CPU depends on that median, so the value check is circular unless an external value anchor or a supply bound is supplied.","section":"Section 3.1, design principle"}],"minor_comments":[{"comment":"The displayed 'geometric series' equality uses an infinite product symbol where a sum is intended; it should read n/p + xn/p^2 + ... = (n/p) * sum_{k=0}^{∞} (x/p)^k, not an infinite product.","section":"Section 3.2, Proposition proof"},{"comment":"The notation '5c2p' is confusing and should be written as 5 c^2 p, with an explicit explanation of why the constant 5 is chosen.","section":"Section 4.2, Equation (1)"},{"comment":"The word 'comparsion' should be 'comparison.'","section":"Section 5.1"},{"comment":"Reference [9] contains a duplicated 'that that' in its title; please fix.","section":"Reference [9]"},{"comment":"Figure 2 mixes token flows, control flows, and monitor roles in a dense diagram; consider separating these layers or labeling the token types directly on each edge for readability.","section":"Figure 2"}],"recommendation":"major_revision","confidential_remarks":"The paper reads as a protocol design note rather than a complete economic analysis. I would ask the editor to require the authors to either derive the x < p condition from explicit models of conversion volume and strategy choice, or to clearly reframe the sustainability claim as a conjecture. The exchange-market Claim should likewise be restated with r as an endogenous median price. The phrase 'reasonable conditions' should not carry the weight of a theorem at the paper's core."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core idea is genuinely neat: fix task payments in a CPU token that is stable by definition—one token always pays for one computation step—and push all price volatility onto TRU, a minted reward token. That part works by construction, and the median-local-price mechanism with staking monitors is a real design contribution. The paper is honest that it's an outline; the DAO dissolution idea, where governance tokens convert into utility tokens and then vanish, is also interesting and fits the decentralization goal. The exchange arbitrage claim, for a fixed conversion rate, is standard and fine.\n\nThe soft spot is the sustainability Proposition in Section 3.2. The assumptions—that at least p fraction of Solvers/Verifiers hodl TRU (Strategy 2) and that new CPU creation per epoch stays below p—are asserted as 'reasonable conditions' but not derived from any individual rationality argument. And the stress-test feedback is real: Strategy 2 posts prices below the median, which pulls the median down. Since the tasking conversion mints CPU = TRU_burned / p_med, a lower median mints more CPU per TRU, so the creation rate x rises exactly when the Strategy-2 coalition succeeds, and can push x above p. The Proposition ignores this feedback loop. There's also a typo in the geometric series (product instead of sum), and the exchange Claim assumes a fixed rate r while r is the endogenous median price. These are addressable gaps, but they mean the sustainability claim is currently unsupported.\n\nWho's this for? People working on token mechanisms for decentralized computation and on stablecoin-style designs without oracles. It's a useful design sketch, not a formal result. I'd send it to a refereed venue if that venue tolerates speculative protocol design, but the referee should push the authors to either prove or explicitly bracket the Proposition's assumptions, and to discuss the median-price feedback. As it stands, I would not treat the economic sustainability as established.","headline":"A clever token-engineering design for Truebit with a genuinely neat stability idea, but the sustainability theorem's assumptions are not derived and may be undercut by the protocol's own pricing dynamics.","tokens_in":12036,"tokens_out":4904,"would_cite":true,"duration_ms":51716,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"In Truebit's model, one CPU token always pays for one computation step.","keywords":["Truebit","token model","stable task pricing","median local price","mintable rewards","governance token","two-token system"],"falsifier":"Track, in a live or simulated Truebit deployment, the fraction of TRU rewards that participants convert to CPU within the pricing-bonding period and the average number of new tokens minted per epoch of tasks: if the converting fraction exceeds the fraction of hodlers or the per-epoch creation rate $x$ is not below $p$, the Proposition's predicted convergence to all tokens held by Strategy 2 participants will fail. Separately, if both TRU and CPU trade on exchanges, persistent deviations of USD(CPU) from $r\\cdot\\mathrm{USD(TRU)}$ would falsify the arbitrage Claim.","tokens_in":1588,"feed_emoji":"🪙","tokens_out":4621,"duration_ms":98196,"temperature":0.7,"pith_summary":"This paper proposes a token economy for the Truebit computation network in which a single CPU token always pays for one computation step, making task prices fixed for task givers. Solvers and verifiers are rewarded with newly minted TRU tokens issued at the median of participants' local prices, so they absorb market fluctuations instead. The authors argue this mintable two-token design delivers stable task pricing, sustainable economics, and fair market pricing without external price oracles, exchanges, or privileged nodes. The same lifecycle lets a governance token, DAO, convert permanently into utility tokens, nudging the network toward autonomous decentralization.","feed_headline":"One CPU token buys one computation step in Truebit's plan","feed_subtitle":"Minted TRU rewards at median local prices let task givers pay fixed costs while solvers absorb volatility.","key_machinery":"The central mechanism is the two-token mintable format: CPU, the tasking token whose value is hard-wired to one computation step, and TRU, the staking and reward token minted at the median local price. The median local price is maintained by staking swaps, in which monitors can exchange tokens against a participant's stake at their declared price minus a fee, and by a tasking conversion contract that burns TRU or whitelisted external tokens to mint CPU at the median rate. The governance token DAO completes the system: each DAO token can be converted once into TRU or CPU, with a back-loaded bonus $f(p,c)=(p+5c^{2}p)N$ that rewards long-term holding, and conversion shrinks governance power until the political layer dissolves.","core_discovery":"The central claim is that decoupling payment from reward through two linked tokens stabilizes task pricing: task givers pay in CPU, whose denominated value is fixed at one computation step, while solvers and verifiers stake and are paid in TRU, minted on the fly at the median of all bonded local prices. Because median pricing governs both reward minting and TRU-to-CPU conversion, no external price feed is needed, and staking monitors punish outlier prices by swapping against the offending stake. The paper's Proposition asserts that if at least a fraction $p$ of solvers and verifiers follow the hodling Strategy 2 and fewer than $p$ new tokens are created per epoch of $n$ tasks, then the hodlers will come to hold all tokens after $n/(p-x)$ tasks, giving TRU value growth relative to CPU. A further Claim holds that if both tokens trade on exchanges, arbitrage between conversion and trading forces the exchange rate $\\mathrm{USD(CPU)}$ toward $r\\cdot\\mathrm{USD(TRU)}$, where $r$ is the tasking conversion rate. Minting rewards also removes the finite jackpot repository as a bound on the largest secure computation, so in theory tasks of any size can be rewarded.","pith_inferences":["An implicit, testable consequence is that the median local price should track the real fiat cost of computation; if it drifts far from hardware-and-electricity costs, the staking-swap arbitrage may be too weak to correct it because the 20% fee and 24-hour bonding delay let outlier prices persist.","The stability guarantee is relative to TRU, not fiat: external exchange prices can still move, so 'one CPU pays one step' holds for task issuance only while arbitrage keeps CPU's fiat value near $r$ times TRU's.","Varying the arbitrary constant 5 in the DAO conversion formula would change how quickly governance dissolves; a simulation of conversion timing could show whether the back-loaded bonus rewards early hodlers at the expense of late converters."],"forward_implications":["Task givers can issue tasks at any time without worrying about token price changes, because holding CPU guarantees the same purchasing power in computation steps.","Solvers and verifiers become the risk bearers: their TRU rewards fluctuate in purchasing power, and median pricing prevents individual participants from gaming the conversion rate upward.","If the Proposition's conditions hold, hodlers following Strategy 2 will eventually accumulate the entire token supply, making TRU appreciate relative to CPU and attracting task givers through lower fiat-equivalent prices.","Minting TRU on demand lifts the old jackpot-repository cap on secure computation size, so the protocol can in principle reward arbitrarily large tasks.","A one-time conversion of DAO tokens into TRU or CPU, combined with the upgrade game that re-mints CPU' and TRU' in new contracts, provides a path from centralized governance to a fully decentralized, upgradable network."],"supporting_citations":[{"why":"Defines the Truebit protocol, verification game, and underpricing attack that the token model must fix.","marker":"[23]"},{"why":"GasToken supplies the precedent of decoupling task payment from reward through a two-token construction.","marker":"[13]"},{"why":"Dai is the reference point for a stable token, contrasted with Truebit's stability tied to computation rather than USD.","marker":"[2]"},{"why":"Tether represents the fiat-pegged stablecoin approach whose assumptions the paper avoids.","marker":"[6]"},{"why":"Uniswap is cited as a source of TRU liquidity for tasking conversion participants.","marker":"[5]"},{"why":"The securities-law framework underlies the claim that apolitical reward distribution may avoid investment-contract status.","marker":"[12]"}],"fun_headline_variants":["Truebit's twin tokens dodge exchange price feeds","Median local prices mint stable Truebit task rewards","Decoupling payment from reward stabilizes Truebit tasks","Minted TRU at median local prices removes price oracles"],"cache_read_input_tokens":14080,"weakest_assumption_plain":"The sustainability argument collapses if a large enough share of solvers and verifiers do not actually choose to hold their TRU rewards, or if new tokens are created too quickly; the paper assumes these 'reasonable conditions' rather than proving they are the rational equilibrium.","fun_headline_variants_meta":{"raw":{"variants":["Truebit's twin tokens dodge exchange price feeds","Median local prices mint stable Truebit task rewards","Decoupling payment from reward stabilizes Truebit tasks","Minted TRU at median local prices removes price oracles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000351,"raw_usage":{"total_tokens":1878,"prompt_tokens":869,"completion_tokens":1009,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":485,"completion_tokens_details":{"reasoning_tokens":945}},"tokens_in":485,"tokens_out":1009,"duration_ms":11669,"temperature":1.0,"reasoning_tokens":945,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:29:29.863580+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Track, in a live or simulated Truebit deployment, the fraction of TRU rewards that participants convert to CPU within the pricing-bonding period and the average number of new tokens minted per epoch of tasks: if the converting fraction exceeds the fraction of hodlers or the per-epoch creation rate $x$ is not below $p$, the Proposition's predicted convergence to all tokens held by Strategy 2 participants will fail. Separately, if both TRU and CPU trade on exchanges, persistent deviations of USD(CPU) from $r\\cdot\\mathrm{USD(TRU)}$ would falsify the arbitrage Claim.","supporting_citations":[{"cited_title":"A scalable ve rification so- lution for blockchains","cited_arxiv_id":null,"evidence_quote":"Defines the Truebit protocol, verification game, and underpricing attack that the token model must fix."},{"cited_title":"Gastoken.io – cheaper Ethereum transactions, tod ay","cited_arxiv_id":null,"evidence_quote":"GasToken supplies the precedent of decoupling task payment from reward through a two-token construction."},{"cited_title":"https://makerdao.com/en/whitepaper","cited_arxiv_id":null,"evidence_quote":"Dai is the reference point for a stable token, contrasted with Truebit's stability tied to computation rather than USD."},{"cited_title":"https://tether","cited_arxiv_id":null,"evidence_quote":"Tether represents the fiat-pegged stablecoin approach whose assumptions the paper avoids."},{"cited_title":"https://uniswap.io/","cited_arxiv_id":null,"evidence_quote":"Uniswap is cited as a source of TRU liquidity for tasking conversion participants."},{"cited_title":"Frame- work for ‘investment contract’ analysis of digital as- sets","cited_arxiv_id":null,"evidence_quote":"The securities-law framework underlies the claim that apolitical reward distribution may avoid investment-contract status."}],"review_version":1}